Negative predictive value of S100B in all types of traumatic brain injury in different aging groups

Abstract

Purpose

Traumatic brain injury (TBI) represents a major contributor to global morbidity and mortality, and the optimization of diagnostic approaches continues to be a matter of considerable scientific debate. The S100 calcium-binding protein B (S100B) value is characterized by a high diagnostic negative predictive value (NPV) and is obtained without exporure to radiation. The aim of the study was to investigate the NPV of the S100B level in the patient cohort aged 75 and older with TBI and compare the results with a cohort of individuals below 75. The hypothesis was, that the S100B value would have a sufficiently high NPV in both patient groups, thereby serving as a diagnostic marker, but with a higher NPV in the below 75 age group.

Materials and methods

A retrospective study was conducted on 815 TBI patients from a Level I trauma center from April 2016 to May 2024. Both, S100B levels and CT scans were obtained within 30 min to 6 h post-trauma. Patients were divided into two groups: below 75 years and 75 years and older. S100B levels ≥ 0.105 µ/L were considered positive.NPV and sensitivity were calculated for both groups.

Results

Among the 815 patients, 76 had normal S100B and CT results, 13 had abnormal CT but normal S100B, 65 had elevated S100B and abnormal CT, and 661 had elevated S100B with normal CT. The overall NPV was 85.4% (95% CI 0.753, 0.909; p < 0.01), with a sensitivity of 83.3% (95% CI 0.776, 0.919; p < 0.01). In the under-75 group (574 patients), the NPV was 86.8% (95% CI 0.816, 1; p < 0.01), and sensitivity was 76.2% (95% CI 0.5, 1; p < 0.01). In the over-75 group (241 patients), the NPV was 77% (95% CI 0.622, 0.878; p < 0.01), with sensitivity of 91.7% (95% CI 0.786, 0.938; p < 0.01). The intervention rate was 0.3% in the under-75 group and 1.6% in the over-75 group.

Conclusion

We found a clear and significant correlation between a negative S100B level and normal CT scan. We believe that the determination of S100B levels significantly reduces the radiation exposure for TBI patients, especially in younger patients. However, its reliability varies by age, warranting further investigation in diverse populations and TBI severities.

Introduction

Traumatic brain injury (TBI) is caused by a trauma, give rise to an diverse array of symptoms and represents one of the most prevalent causes of morbidity and mortality worldwide [1–4]. In Austria, the prevalence of TBI stands at 303 per 100,000 individuals, accompanied by a case fatality rate (CFR) of 3.6% and a mortality rate (MR) of 11 per 100,000, aligning with figures observed in other identified European nations [5]. The global incidence rates for TBIs vary significantly depending on age. While the incidence is 149/100,000 for individuals aged 40–44, it rises to 1213/100,000 for those aged 85–89 [5]. The standard diagnostic tool for the acute assessment of such injuries is computed tomography (CT), which involves the use of radiation [6].

As an alternative diagnostic tool, serum markers can be utilized. The ideal marker should exhibit high specificity for brain tissue and high sensitivity for detecting brain damage [7]. In recent years, multiple biomarkers have been investigated and the S100B marker has been established as a promising biomarker-based strategy for the diagnosis of TBI [8]. – [9] S100B is a calcium-binding protein with low affinity, expressed in Schwann and glial cells, and is released during astroglial injury [9, 10]. Assessing the S100B level in the bloodstream can assist in diagnosing TBI and gauging the injury’s severity. Certain investigations propose that the S100B level could function as a decision-making tool regarding the necessity of CT scanning [11–14]. Already, there is widespread utilization of this serum marker in specific clinical scenarios. Guidelines, such as those established by the Scandinavian Neurotrauma Committee, (SNC) already incorporate this blood parameter as a diagnostic tool for TBI, alongside which additional scoring systems, such as the Glasgow Coma Scale (GCS), are recommended [15]. A more standardized application of this measure could potentially reduce the need for almost every third CT [16–19].

To determine whether this tool is suitable as a diagnostic tool, a statistical value, namely the negative predictive value (NPV), can be employed. The NPV is defined as the probability that a negative test result accurately indicates the absence of a disease or injury. If the NPV of S100B is exhibits high rates, it could potentially serve as a diagnostic alternative to CT scans, helping to avoid unnecessary radiation exposure and costs. However, there is limited scientific evidence, particularly in the context of CT examinations in older patients, regarding the NPV of S100B. A high NPV of the S100B value, combined with its high sensitivity, can further enhance its diagnostic value. It should be noted that this value can be determined relatively quickly through a blood sample, compared with the considerably more resource-intensive CT scan. Especially in the very old population clinical assessment and CT decission might be difficult in the presence of dementia, intake of anticoagulants [17–19].

The aim of the study was to investigate the NPV of the S100B level in the patient cohort aged 75 and older with TBIs and compare the results with the cohort of individuals below 75. The hypothesis posited that the S100B value would have a sufficiently high NPV in both patient groups, thereby serving as a diagnostic marker, but with a higher NPV in the below 75 age group. The null hypothesis suggested that the NPV remains consistent regardless of age.

Materials and methods

Study design

A retrospective single-center study aimed to explore the correlation between the NPV of the S100B value and the identification of clinically significant traumatic brain injuries. The research spanned from April 2016 to May 2024, encompassing a total of 815 cases during that specific timeframe.

Participants

All patients, which were admitted to the emergency department of a level I trauma center exhibited TBIs were included, whereby all severity grades were included. Both CT scans and S100B level determinations were mandatory conducted for all patients to enable a comparative analysis of these outcomes. Both, CT and S100B evaluation were performed on the day of admission. It is crucial to note that only S100B values from blood samples processed between 30 min and 6 h post-trauma were included in the analysis.

Inclusion and exclusion criteria

Patients were included in this study if they were admitted to the Level I trauma center with a TBI during the specified period and had both, a documented CT scan and a valid S100B measurement. To be eligible, patients had to be at least 18 years old at the time of hospital admission.

Patients were excluded if they were admitted outside the study period, lacked sufficient documentation of the S100B measurement or CT scan, or did not meet the minimum age requirement. Furthermore, pregnant patients were excluded from the study.

Data collection

Patient data were extracted from the Hospital Information System (HIS), including demographics, S100B values, CT findings, GCS scores, and information on surgical interventions. Data extraction was independently performed by two reviewers to ensure accuracy and consistency.

Intervention

The determination of the S100B value involved obtaining blood samples using an EDTA tube. Subsequently, this blood sample was dispatched to the hospital laboratory for analysis by a machine. It is crucial to note that blood samples were eligible for analysis only if collected within the timeframe of 30 min to 6 h post-trauma. The threshold designating a positive S100B value was established at 0.105 ug/L. This value was assessed upon arrival in the emergency room. Furthermore, upon arrival at the emergency department, a cranial CT scan was performed and evaluated by a board-certified radiologist. Intracranial and extracranial pathologies (hemorrhage, fracture) were documented, with extracranial fractures involving the viscerocranium, excluding isolated nasal bone fractures.

The GCS score was assessed by a specialist in traumatology and subsequently categorized according to international guidelines as follows: GCS 15–13, mild traumatic brain injury; GCS 12–9, moderate traumatic brain injury; and GCS 8–3, severe traumatic brain injury.

All surgical procedures directly related to the TBI case and involving the neurocranium (craniotomy, placement of an intracranial pressure probe, intubation of the participant, and other unspecified interventions) were identified in the HIS. These procedures were performed by board-certified neurosurgeons.

Outcome

The primary outcome of this study was the NPV of serum S100B levels for detecting traumatic brain injury in the overall cohort as well as in predefined subgroups. Subgroup analyses were stratified by age (< 75 years and ≥ 75 years) and by the severity TBI as classified by the GCS.

Secondary outcomes included demographic characteristics (age and sex distribution), clinical parameters including the GCS score at admission, and the presence of neurocranial and viscerocranial fractures as identified on cCT. Additionally, data on surgical interventions directly related to TBI (craniotomy, intracranial pressure monitoring, and other neurosurgical procedures) were evaluated.

Statistical analysis

The dataset was compiled into a spreadsheet using Excel^® (Microsoft^® Corporation, Redmond, WA, USA) for statistical evaluation. Data analysis was subsequently performed with IBM^® SPSS^® (Statistics 29.0, Armonk, North Castle, NY, USA). The patient findings from the CT scans and S100B values were extracted from the HIS. All data were scanned by two independent individuals to minimize errors. Two distinct age groups were analyzed: patients below 75 years and those 75 years and older. The S100B values were coded as 0 or 1, with 0 indicating a normal S100B level and 1 indicating an elevated level. Similarly, the CT findings were coded, with 0 representing a normal CT result and 1 indicating an abnormal finding consistent with intracranial hemorrhage (ICH). For each group, the NPV and p-value were analyzed separately. Statistical significance was set at a p-value < 0,05. Post hoc power analysis according to Hoenig and Heisey was calculated according to the magnitude of differences of our main endpoints and revealed that the included number of cases (815) was sufficient to reveal a power greater than 80% with a p-value less than 0,05. Furthermore, the following interventions were also extracted from the HIS by two different individuals and analyzed by subgroup: craniotomy, placement of an intracranial pressure probe, intubation of the participant, and other unspecified interventions. Confidence intervals (95% CI) were calculated to estimate the precision of the effect size measures.

Results

Age and gender

A total of 815 TBI cases of a Level I Trauma center with valid S100B measurements and CT findings were identified. Among these individuals, the mean age was 49.4 years (median: 50 years), with an overall age range spanning from 18 to 100 years. When analyzed by gender, the average age in the female cohort was 55 years, with a median age of 43 years and an age range of 18 to 98 years, while in the male cohort, the mean age was 49.4 years, with a median age of 43 years and an age range of 18 to 98 years. (Fig. 1)

Fig. 1.

Density of patient age by gender

When analyzing the group of individuals below 75 years of age, there were 274 women and 296 men, with a mean age of 38.8 years, a median age of 34 years, and an age range of 18 to 74 years.

The group aged 75 years and older consisted of 145 women and 89 men, with a mean age of 85.4 years, a median age of 85 years, and an age range of 75 to 100 years. (Fig. 2)

Fig. 2.

Age distribution by gender in age group

CT and S100B findings of the entire cohort

Among the entire cohort of 815 TBI cases, 76 cases were found to have both negative S100B and CT results, 13 cases showed pathological CT findings but physiological S100B values, 65 cases had elevated S100B and pathological CT findings, and 661 cases showed elevated S100B levels but physiological CT findings. This results in a NPV of 85.4% (95% CI 0.753, 0.909; p < 0.01), a specifity of 10.3% (95% CI 0.069, 0.112; p < 0.01) and a sensitivity of 83.3% (95% CI 0.776, 0.919; p < 0.01). Fractures were evident in 17 cases, of which 6 (35%) were extracranial and 11 (65%) intracranial in location. (Fig. 3)

Fig. 3.

Confusion matrix comparing S100B levels with CT findings in TBI patients of the entire cohort

CT and S100B findings of the cohort “below 75 years old”

In the analysis of the cohort below 75 years old, a total of 574 cases were identified. Of these, 66 patients had both physiological S100B values and normal CT findings, 10 had pathological CT findings but physiological S100B values, 32 showed both pathological CT and elevated S100B values, and 466 had elevated S100B values despite physiological CT findings. This results in a NPV of 86,8% (95% CI 0.816, 1; p < 0.01), a specifity of 12.4% (95% CI 0.1, 0.189; p < 0.01) and a sensitivity of 76.2% (95% CI 0.5, 1; p < 0.01). Fractures were evident in 6 cases, of which 2 (33%) were extracranial and 4 (67%) intracranial in location. (Fig. 4)

Fig. 4.

Confusion matrix comparing S100B levels with CT findings in TBI patients of the cohort “below 75 years old”

CT and S100B findings of the cohort “75 years and older”

In contrast, the cohort of patients aged 75 years and older comprised 241 cases. Of these, 10 patients exhibited both physiological CT findings and S100B values, 3 had physiological S100B values but pathological CT findings, 33 had both pathological CT findings and elevated S100B values, and 195 had elevated S100B values despite physiological CT findings. This results in a NPV of 77% (95% CI 0.622, 0.878; p < 0.01), a specifity of 4.9% (95% CI 0.044, 0.085; p < 0.01) and a sensitivity of 91.7% (95% CI 0.786, 0.938; p < 0.01). Fractures were evident in 11 cases, of which 5 (45%) were extracranial and 6 (55%) intracranial in location. (Figures 5 and 6)

Fig. 5.

Confusion matrix comparing S100B levels with CT findings in TBI patients of the cohort “75 years and older”

Fig. 6.

Distribution of TBI cases by case categories

GCS findings of the entire cohort

Of the 815 patients included, the vast majority presented with a GCS score of 15 (n = 748), followed by GCS 14 (n = 26) and GCS 3 (n = 16). Lower scores were comparatively rare, with GCS 13 (n = 8), GCS 11 (n = 4), GCS 6 (n = 4), GCS 7 (n = 3), GCS 10 (n = 2), and GCS 8 (n = 2). Single cases were observed with GCS 12 and 5 respectively.

When analyzed by severity of TBI, the majority of patients (n = 782; 95.9%) sustained a mild injury, followed by severe TBI in 30 cases (3.7%), while only three cases (0.4%) were classified as moderate.

Interventions

The group of individuals below 75 years old showed a total of 2 interventions (0.3%), both of which were craniotomies. In contrast, the group of individuals over 75 years old showed 4 interventions (1.6%), namely 1 craniotomy, 2 intracranial pressure probes, and 1 intubation. The intervention rate for the entire population is 6 interventions out of 815 patients (0.7%). (Fig. 7)

Fig. 7.

Interventions performed by age group

Discussion

The aim of the study was to investigate the NPV of the S100B level in the patient cohort aged 75 and older with TBIs and compare the results with the cohort of individuals below 75.

The hypothesis was, that the S100B value would have a sufficiently high NPV in both patient groups, thereby serving as a diagnostic marker, but with a higher NPV in the below 75 age group.

The entire study population of 815 TBI patients demonstrated a relatively high NPV of 85.4% for the S100B value, suggesting the potential consideration of S100B as a diagnostic tool. Comparing the two age groups confirmed the study hypothesis, as the younger age group exhibited a significantly higher NPV of 86.8% compared to the NPV of 77% observed in the 75 years and older group.

The study results clearly demonstrated that the S100B biomarker should be involved in the diagnosis of TBIs, especially considering that this diagnostic tool only requires a blood sample. However, it should be noted that the marker’s reliability is influenced by several intrinsic and extrinsic factors [20]. According to Thelin et al. [21], the timing of the sample collection is an essential factor for obtaining valid results. Lange et al. [22]. demonstrated that in sober patients, the S100B value has significantly higher reliability compared to patients with recent alcohol consumption. Furthermore, they showed a correlation between the reliability of the S100B and the patient’s age, with specificity decreasing as age increases [22]. Furthermore, the ethnic background should also to be mentioned, as it also influences the validity [23].

Since the S100B value is one of the most well-studied biomarkers for TBIs, it has already often been cited as promising [20]. Not only because of its high NPV, but also due to cost savings for the healthcare system, the S100B biomarker has become popular [24]. For example, a cost saving of up to 71 euros per patient with mild TBIs, characterized by a GCS of 15 − 14, has been demonstrated [24].

In other countries, such as Scandinavia, the inclusion of the S100B marker in diagnostics is already evident, published by the SNC. There, patients with mild to moderate TBI and a GCS not lower than 14, if a physiological S100B value is observed within 6 h post-trauma and no additional risk factors are present, an initial CT scan is not performed, and the patients can be discharged from the hospital [25, 26].

However, there is a trend in the literature towards including only mild TBIs when considering the S100B value, which results in a comparatively higher NPV if compared to this study, in which all admitted TBIs regardless of severity were incorporated. Steinmüller et al. [27] achieved an NPV of 99% by including only mild TBIs, which correlates with other study findings. However, they did not differentiate by specific age groups.

An important aspect of a diagnostic tool is not only a high NPV but also the prevention of a potentially avoidable dangerous course, such as an undetected ICH in TBI. The study population shows that in only 0.7% of cases, an intervention was necessary to prevent potentially life-threatening outcomes. Interestingly, the group below 75 years old had a more than five times lower intervention rate compared to those over 75, with 0.3% vs. 1.6%. It could also be emphasized that determining the S100B value is especially effective and desirable in younger patients.

Another point warranting discussion in this study is the notably low specificity of the S100B value, only 10.3% for the overall cohort, and 12.4% and 4.9% for participants younger than and equal to/older than 75 years, respectively. Current evidence aligns with our findings, indicating that despite the very high NPV and sensitivity, the specificity remains comparatively low. Karamian et al. [12] demonstrated in their meta-analysis a specificity of 32% for S100B, likewise accompanied by high NPV and sensitivity. Similarly, Santing et al. [13] examined S100B levels in individuals aged over 65 years and concluded that specificity declines markedly with increasing age, from 22% in those aged 65–74 years to merely 5% among individuals older than 85 years, findings that are consistent with the results of the present study. Furthermore, it is crucial to note that this study detected intracranial or extracranial fractures in 2% of the study population, a factor that directly influences the specificity of the S100B marker [12–14].

A controversial topic is the diagnosis of TBI and the use of the S100B marker in older adults, not least because elderly individuals carry a higher risk of intracranial pathologies and often present with polypharmacy and multiple comorbidities. Consequently, advanced age alone is one of the main indications for cCT, although most scans reveal no pathological findings [28]. However, there is sufficient evidence that S100B, in combination with clinical scoring systems, can be safely applied even in older populations, a conclusion supported by the present study [13, 18]. Oris et al. [18] investigated S100B levels in elderly patients and concluded that, under optimized conditions, S100B can serve as a diagnostic tool in individuals over 80 years of age, specifically when the threshold is raised to 0.15 µg/L. This adjustment allowed 29.3% of cCT scans to be avoided in this cohort.

Limitations

This study has several limitations, starting with the fact that the entire study population was collected at only one Level 1 trauma center. We decided to consider the S100B value as negative if it falls below < 0.10 µg/L, as this reference value is also cited in most studies [27–31]. However, a certain number of subclinical, undetected brain injuries or long-term damages cannot be definitively ruled out, as there was no follow up conducted. In this study, ethnic background was not differentiated, which is, according to the study by Abdesselam et al. [23], also a factor of influence. However, the majority of the patients included are Caucasian, reflecting the average population of Austria. Furthermore, the sample size of 815 TBI cases should also be noted, as it may limit the statistical power of the analysis. The interpretability of our findings may be further limited due to the lack of control for patients’ medications and comorbidities.

Conclusion

We found a clear and significant correlation between a negative S100B level and normal CT scan. We believe that the determination of S100B levels significantly reduces the radiation exposure for TBI patients, especially in younger patients. However, its reliability varies by age, warranting further investigation in diverse populations and TBI severities.

Abbreviations

CFR

Case fatality rate

CI

Confidence intervall

CT

Computed tomography

GCS

Glasgow Coma Scale

HIS

Hospital information system

ICH

Intracranial hemorrhage

L

Litre

MR

Mortality rate

NPV

Negative predictive value

SNC

Scandinavian Neurotrauma Committee

TBI

Traumatic brain injury

µg

Microgram

Author contributions

C.C.: Conceptualization; methodology; investigation; writing—original draft preparation. P.P: Conceptualization; methodology; analysis of data; investigation; data curation; supervision; writing—review and editing. S.V.: Conceptualization; methodology; analysis of data; investigation; data curation; supervision; writing—review and editing. D.K.: Conceptualization; methodology; analysis of data; investigation; data curation; supervision; writing—review and editing. A.L.: Conceptualization; methodology; analysis of data; investigation; data curation. P.S.: Conceptualization; methodology; analysis of data; investigation; data curation. P.R: Conceptualization; methodology; analysis of data; investigation; data curation; supervision; writing—review and editing.

Data availability

The data that support the findings of this study are not openly available due to reasons of sensitivity but are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.